[Michael Mccune]

Interesting to compare a tracking attempt in Resolve and then in Fusion.

Example: try something very simple, such as a small black microphone against a white shirt and light background. No foreground objects. The goal is to track and capture some minor panning motion for later use with other power windows.

Run the tracker (probably the cloud version) with a small power window in Resolve, noting the time (speed) and the general success in sticking with the microphone.

Then run the same video in Fusion with its tracker. Sometimes a huge positive difference both in speed and success in sticking with the object. What is your impression? Probably Fusion is found to be much faster and more reliable.

But Resolve has a much better approach to correcting errors, allowing one to use Frame mode and create intermediate correcting and adjustment keyframes, adjust the polygon shape as needed, delete and interpolate mis-tracking segments, and so on. Very easy and effective. Too bad there is so much more to correct than with Fusion’s tracker output.

Fusion, comparatively, is much more difficult to correct and tweak, (probably just my lack of understanding how it really works). Fusion has a keyframe mode but adding keyframes, deleting them, moving them, etc., is not obvious to me (yet). And interpolation between deleted keyframes seems to work differently in Fusion (?).

Otherwise, I would use Fusion because it is so much faster and more reliable in sticking with the object.

Someday I am sure BMD will get the best of both worlds into one method. Would love to import the tracking data–in the manner of importing a Fusion alpha into Resolve via External Mattes–and then use attach that Fusion tracking data (instead of the Resolve tracking data) to Resolve power windows. Ahh, someday …..

Mike

[Michael Mccune]

Marc; yes, you were correct in pointing to the link to the gallery folder.

Gallery file corruption actually seems to have been main culprit. In fact, the next morning the entire drive was not recognized by Windows. I use Hard Drive Sentinel to monitor drives and it said it was fine though windows could not read it (???). In fact the *drive* was fine but it failed trying to run some read/write tests.

Short story, most all files could be copied off that drive except the gallery folder, some files of which could not be read. This may explain Resolve showing gallery stills though the wipe function did not work (black replacing any gallery image).

So, rebuilding and re-linking the gallery folder seems to be the fix. We will know soon!

Thanks again, Marc.
MIke

[Michael Mccune]

Simon; are you saying that the Planar Tracker works as many tracking points, not just, for example, the four corner points you used in your Planar Tracking demonstration???

https://youtu.be/FyOMddTEGmw

That is, how do you set up the bear polyline and then try to track that polyline???

As we know, Resolve allows multi-point tracking and “Cloud” tracking. But does not track a polyline such as a mask.

But is there hope …. that the Planar Tracker can be coaxed into a pseudo roto tool???

Thanks, Mike

[Michael Mccune]

… It was the Colosseum, not the city of Rome, that was built in a few years. But they did have 100,000 slaves.

Mike

[Michael Mccune]

Right; Rome wasn’t built in a day; it was indeed built in record time, a few years, but not just a day. (They didn’t use mortar between the blocks but iron clamps and they had 100,000 slaves.)

I am sure that BMD and others will, in due course, have Photoshop-like marching ants masking tools that edge-detect with a level of accuracy that would revolutionize rotoscope work.

All we need are 100,000 slave engineers.

Thanks, Mike

[Michael Mccune]

Vaguely recollect this error, perhaps in Avid; may have been a corrupt frame in a frame sequence source.

As I recollect, found it by trial and error, that is, dividing the clip into two, test each segment to find the segment still producing the error; rinse and repeat; … finally reducing it to a segment size small enough to scrub frame by frame. Then look for a single bad frame.

The culprit even be identified and flagged at the frame level as corrupt.

… just a thought.

Mike

[Michael Mccune]

Glenn; yes, that indeed was the fix: remove Fusion 8.

Only by chance I saw the splash screen say, “Fusion 8” when of course I expected “9.”

Don’t recall any Fusion install prompt about removing the previous version.

So I probably misled myself as Resolve seemed to allow multiple installed versions.

Anyway, all’s well that end’s well.

Thanks again, Mike

[Michael Mccune]

Thanks, Simon; here is a delayed reply.

The overall goal probably seems rather simple: create multiple mattes within a clip for color adjustments.

But multiple Resolve power windows and the associated tracking slow down the PC markedly, eventually becoming unworkable. I have used perhaps seven or eight masks with tracking and a total of over three or four dozen nodes: it all works but sooo slowly that it eventually fails, even with dual Xeons and 48 GB RAM and a GeForce 1080 series card.

Fusion to the rescue: perhaps Fusion’s overall compositing skills can help. Especially with multiple masks that need to be combined. Indeed I seem to find that Fusion’s ChromaKey is actually quicker and more reliable than the several filter options in Resolve, such as HSL, Luminance, 3D or RGB, even with the various refinements applied.

Then Fusion can offload this work and its exported masks are imported into Resolve as external mattes. Works fine. Especially as Fusion Booleans can export RGB as black plus the alpha channel, keeping file size down, usually to less than a Megabyte per frame.

Indeed, Resolve seems to handle over half a dozen simultaneous external mattes without the slowdown from Resolve’s tracking, windows and filters.

Not sure if this multiple external matte process can be replicated within Fusion Connect. But it would be a huge help to have an active FC connection. Then perhaps tweak the mask in Fusion and have it recognized in Resolve on the fly.

So that is my overall plan for now. Not perfect but might get me out of the woods.

Thanks for all your help, Simon.

Mike

[Michael Mccune]

Simon; I very much look forward to it. I have actually studied, yes, studied your tutorials. That is, repeat sections trying to catch the quick but necessary details that are part of some process or other. And try to understand why my attempts simply don’t work.

Seems very often that the answer is that there are “modes” or preliminary setup or context options that must be set before the process at hand is attempted. Then things work as expected.

The joke I tell friends, eventually, is that the secret is that one must click one’s heels and turn around twice. Then it works. But that is the same with any very complex system, I would say, where the technician’s technical expertise is, as they say, indistinguishable from magic!

Again, I look forward to your tutorials very much.

Thanks, Mike

[Michael Mccune]

Thanks, Simon. Yes, I had not thought about the Prefs file.

Did not however find one in Windows Fusion: searched through every Fusion folder.

The file “com.blackmagic-design.fusion.plist” is on your Mac, right?

The good news is that deleting all the defaults names and file paths wherever I found them and entering new ones–in Fusion Preferences or in Save dialog boxes–seems to have restored Save and Save As so they now succeed as expected.

Thanks again, Simon.

P.S. Have you ever thought about a Lynda.com course on Fusion??? Is there a market??? Resolve has a few wonderfully complete–that is, beyond basic to include intermediate use–courses by Pat Inhofer. Not much out there it seems for Fusion.

Seems helpful often to have more than the “tutorial” format–“Do this, then do that, …” The tutorial format provides is one layer of understanding, to be sure.

But also helpful may be information on the “logic” level, say, how boolean syntax and algebra work and show how it is applied to masks and mattes. In computer languages they might call this the “grammar.” As it is, one has to piece it together from Photophop books (such as, “The hidden Power of Blend Modes,” by Scott Valentine) and the internet.

And there is the “cookbook” format: a collection of specific comp tasks and how to accomplish them. As you know, cookbook examples do not go down to the low level stuff nor up to the concept level, but they do provide quick possible roadmaps.

Then the higher level concepts:“why” something is done, the purpose, advantage or strategy vis-a-vis another. An overview, from 30,000 feet, as they say these days.

I suspect that you all, who have mastered video graphics at a very high level, have actually learned all of this though beginners like myself are sometimes helped with a structured approach and visiting all these levels.

Yours, Mike