E̶a̶s̶y̶ Beat Meters with Davinci Resolve / Fusion

[MortonBridges]

So, I've been playing around with Davinci Resolve and stumbled upon this nice way to create a beat meter using Fusion. I don't think I've seen anything like this posted before, either here, on Reddit, or elsewhere and if I'm wrong I humbly ask for forgiveness. Or maybe this is just too dang obvious, in which case, again, I humbly beg forgiveness. This is my first attempt at doing anything with Fusion, so I wouldn't be surprised, but also I'm surprised that my searches for something like this have come up empty. So, for those in the future who use the same search terms as I did, hopefully they'll find this thread.

In addition to the simplicity this method offers extreme customization. You are limited only by your imagination (and artistic skills...)

This uses entirely free-to-use software. I've made my test clip on Linux.

You will need:
A beat making tool, such as Hydrogen - http://hydrogen-music.org/
Audacity - https://www.audacityteam.org/download/
Davinci Resolve - https://www.blackmagicdesign.com/produc ... ciresolve/
Reactor - https://www.steakunderwater.com/wesuckl ... php?t=3067

The Setup

1. Lay out your beats in your preferred beat-making software.
2. Open your beat track in Audacity and follow the steps at https://forum.audacityteam.org/viewtopic.php?t=115891 to create a square-wave-ish representation of the waveform. You'll need to tweak the (setf threshold 0.2) and (setf time-res 0.1) values until you get things right. The exact values will depend on your original beats. You will be looking for a single bar at the peak of your beats. This will result in the new version of the wave form to have values of 0 (or more accurately near-zero) and 1, which is perfect for when we import this into Fusion.
3. Save this new audio as a .WAV file. The Fusion tool to import audio only supports WAV, as far as I can tell.
4. Fire up Davinci Resolve. If you don't have Reactor install it. (https://www.steakunderwater.com/wesuckl ... php?t=3067)
5. Open Reactor (Workplace -> Scripts -> Comp -> Reactor -> Open Reactor). Search for the "Suck Less Audio" tool and select it.

You may need to restart Resolve.

The Fun Part

1. Open Fusion. Add one of each node for particle emitter, particle render, and media output. Link them all up.
2. Select the particle emitter and make the following changes:
   • Region: Line. Drag this to where you want the beats to start. Dragging it off the viewable area is possible, so the beats can 'flow in' from the side. Make this as SMALL AS POSSIBLE to force the particles in a straight line.
   • Style: Change the Style to Brush, and the Brush to "ball_metal"*. Change the size and color to your liking.
   • Controls: Velocity - Set the velocity to a decent speed for your track. Set the angle to 180 (assuming a right-to-left travel).
   • Controls: Emitter - Set the Lifespan such that the particles disappear at the point you want them to stop.
   • Controls: Emitter - Right click on Number, select Modify With -> Audio (WAV). Go to the Modifier tab.
   • Modifiers: Browse for the 'square wave' WAV file.
3. Click the Play button. Things should start flowing. You won't hear the beats, so use your imagination. Adjust size, velocity, lifespan and such to your liking.
4. To make the beats 'pop', on the particle emitter you can use Style -> Size Control -> Size Over Life. This can make the particles change size as they travel. This can be used to give the beats a pulse when they're hit.
5. In the Edit view, add your regular beats track. Line up the first beat in your audio track with where the beats are hit. Moving frame by frame should make this easy.

The rest is all artistic. Add some graphics for the track, hit indicator, and such. You can do this in Fusion, or make an image that your beats will ride over. The possibilities are endless.

Notes:
* - The metal ball is a simple choice, but obviously you can choose any of the available brushes. Or use a "blob" style. Whatever fills your boots. But if nothing catches your eye set the Style to "bitmap". When the style is bitmap the particle emitter will have a new input, to which you can map a Media In node that references the image you want to use. How simple is that?

[MortonBridges]

Very, very simple sample.

https://cloud.degoo.com/share/431EpMCshIVZeM7wgYcjxw

[CantoFan123]

Very, very simple sample.

https://cloud.degoo.com/share/431EpMCshIVZeM7wgYcjxw

Nice!! I just started using resolve for work and if i ever get around to trying my hand at a cock hero it'll be awesome to use this! Thanks for sharing!

[3xTripleXXX]

Cool. I've just dropped the Adobe Suite, since it just got too expensive and I'm going to do the next video with DaVinci Resolve. I've only fiddled with it a little yet (I'm still in the content gathering phase), but it seems promising.

For those of us who use Beatmeter Generator, importing image sequences works brilliantly in Resolve as well, from what I could tell, so that's a super simple solution. :)

However, I'm definitely going to have to get familiar with Fusion for the stuff I used to do with After Effects, so it's great to see some examples of that kind of stuff.

[fragrantEmulsion]

Another feature I used resolve for was automatically splitting a video on scene changes.

[masperturbator]

The Degoo page is loading today, but the video won't play. I'm replying based on the thread and what I searched after.

What I'm taking from this is that Reactor 3 is a plugin manager for Resolve, and it has access to the Suck Less Audio plugin. The Audacity step is needed to simplify a beatmeter audio track into square waveform, because the Resolve Fusion animation needs that absolute contrast to correctly render a meter graphic from the audio file.

Can you give a project export that shows the description?

If I'm right about the reason for the Audacity step, we can solve that very simply by using a correct beat sample file in BMG.



Edit:

At instructions, "Open Reactor (Workplace -> Scripts -> Comp -> Reactor -> Open Reactor)" fails in Resolve 17 where "Comp" leads to nothing responding. May be that Resolve 17 is newer than Reactor 3.

[MortonBridges]

The Degoo page is loading today, but the video won't play. I'm replying based on the thread and what I searched after.

Strange about the Degoo link. It works for me when I connect as anon. Another sample is up on my Mega for my current project (opening round, female and male undressing and masturbation) at https://mega.nz/file/URxmlSCZ#EVtceEJpa ... BLMVgncr7o

What I'm taking from this is that Reactor 3 is a plugin manager for Resolve, and it has access to the Suck Less Audio plugin. The Audacity step is needed to simplify a beatmeter audio track into square waveform, because the Resolve Fusion animation needs that absolute contrast to correctly render a meter graphic from the audio file.

Correct on the Reactor bit.

Suck Less Audio allows you to use a WAV file to supply the data as a to a Particle Emitter number value. The particle emitter, from what I can tell, use the number value as the probability to emit a particle. The 'square' wave allows for a 1 at times where you want the beat (100% probability) or a 0 when you do not (or effectively 0, I don't think Audacity actually exports with a literal 0 value).

This is about the extent of my Fusion knowledge, and there may be better ways, but I've not been able to uncover them. I'm approaching "pre-novice" level of Fusion ability, so I fully expect I'm doing things inefficiently.

Can you give a project export that shows the description?

I'll see what I can do.

If I'm right about the reason for the Audacity step, we can solve that very simply by using a correct beat sample file in BMG.

Thanks. I'll need to look into that (and BMG in general).

Edit:

At instructions, "Open Reactor (Workplace -> Scripts -> Comp -> Reactor -> Open Reactor)" fails in Resolve 17 where "Comp" leads to nothing responding. May be that Resolve 17 is newer than Reactor 3.

I started with Resolve 16, and later upgraded to 17, on both Linux and Windows. Both installed equally well, and Reactor appears to be working in 17, and the Comp menu loads fine on my box.

Now for some Stuff I Found Out™.

Getting the wave form perfect is very, very tricky. I don't know if it's just the resolution of Resolve / Suck Less Audio, or whatever, but sometimes a beat will be "missing", or a new beat may be inserted (usually by 'doubling up' an existing beat). There are tools in the Suck Less Audio modifier that can tweak this (Proxy Sampling), but I don't know the perfect combination of these settings and Audacity Nyquist Prompt options (that's another black box). What I normally do is find the offending beat in Audacity and stretch it (for ones that are missed) or shrink it (for ones that are doubled). Not great. It can be very time consuming (particularly since my machine is woefully underpowered to be doing much Fusion...)

And when beats are "too close" then things just don't look right. I need to look into that...

[masperturbator]

Thank you, I have it working enough to poke at it now.

This video summarizes what's happening here: https://www.youtube.com/watch?v=4nqCx0Aacp8

The OP for Suck Less Audio File Modifier explains how the input WAV file may be interpreted.

Now for some Stuff I Found Out™.

Getting the wave form perfect is very, very tricky. I don't know if it's just the resolution of Resolve / Suck Less Audio, or whatever, but sometimes a beat will be "missing", or a new beat may be inserted (usually by 'doubling up' an existing beat). There are tools in the Suck Less Audio modifier that can tweak this (Proxy Sampling), but I don't know the perfect combination of these settings and Audacity Nyquist Prompt options (that's another black box). What I normally do is find the offending beat in Audacity and stretch it (for ones that are missed) or shrink it (for ones that are doubled). Not great. It can be very time consuming (particularly since my machine is woefully underpowered to be doing much Fusion...)

And when beats are "too close" then things just don't look right. I need to look into that...

The WAV file used to signal beats must be made specially for the mode that suck less is set to. You'll notice that in the square waveform I made off the cuff, I had a positive dB and negative dB square. I've changed that to a single positive dB pulse that I fed to Beatmeter Generator as a custom beat sample.



The BMG output needs to be passed through Audacity to change its encoding to Signed 16-bit PCM, because BMG outputs 32-bit.

I tested a BMG output of 1 minute of 500 BPM. Applying that square waveform to a layout position value (offset from screen center) for a text title node, similar to the video tutorial (but a different value than he used,) is perfectly consistent except that the timeline frame rate will occasionally force a duplicate frame.

When the WAV is silent:



When the WAV is non-silent:



That "silence" also shows there is a small amount of noise in the beat sample WAV, because that value being changed is originally set to "0.500," but it rests at "0.5001525925474"

Also notice:



The waveform I gave it, compared to the modification that Suck Less made to the value it was attached to:



We should probably expect that negative dB values on the waveform will cause negative values on the modifier, and that will come out to be Original Value + negative dB amount. On this example, we should need -0.5 dB wave form to make that value equal 0.

Edit: I have your particle emitter acting as I expected. It definitely needs a precise "1" db from the waveform.

Edit: Watching it react to some changes, and considering other involved variables, this is going to be rather difficult to use for a beatmeter, because the particle line's Velocity value should be the song's BPM, and there we're getting into per-song maths. Also, I did confirm that (linear) 1 pulse on the input WAV file will produce a 1 on the particle emitter's Number value, but the particle emitter isn't consistent enough to do what we're trying here. The particles pop in and out of existence over time, as if there is some intentional randomness still in play.



Suck Less is precise in its input WAV reading. The text node I tested will move exactly as directed by the modifier. The problem you found is in the particle emitter, and I don't see an obvious solution to get the emitter to behave that way. It might be that the particle Number value isn't entirely what we think it is.

This method is, however, perfect for the non-linear meters that we've seen in some Cock Hero videos, where the author is using a graphic that pulses or pops in and out of visibility.

Example: An image node in Fusion, with Suck Less reading a beat file that is produced with a sine-wave beat sample, will correctly animate the Opacity value of that image, so it is only visible during the beat hits.

Example: All of the flashing stroke prompts in Cock Hero No Mercy can be done with what I described above, using a square 0-to-1 waveform.

Edit: https://mega.nz/file/KoVUETwa # geV7Fk8k4pJxMjEFc7FRSY1l_1LsaqklaAwMBHuaP18

That is a zip of both the Resolve project archive and the BMG timeline, from the test I just wrote.

[MortonBridges]

Watching it react to some changes, and considering other involved variables, this is going to be rather difficult to use for a beatmeter, because the particle line's Velocity value should be the song's BPM, and there we're getting into per-song maths.

I'm not so sure about that. I have a 30 minute beat track under way with this technique, with a constant crawl speed, across multiple songs with varying BPM (sometimes varying in the same track, because life isn't hard enough, apparently) and it flows well enough for my liking.

Suck Less is precise in its input WAV reading. The text node I tested will move exactly as directed by the modifier. The problem you found is in the particle emitter, and I don't see an obvious solution to get the emitter to behave that way. It might be that the particle Number value isn't entirely what we think it is.

This is speculation on my part, but I believe that Audacity might not be setting 'silence' to exactly 0dB. If true this will result in double-beats being generated from time to time, since the particle emitter generates particles based on probability, and 'nearly zero' probability is not nearly-zero enough.

The solution I'm using to correct for this is to isolate the offending portion of the beat wave form, save that as a WAV, rerender just that bit until everything looks OK (it usually takes only one pass) and then patch that into the master beat meter. I'm looking for better options, but so far this has given me the best and quickest results.

So maybe it's time for me to eliminate EASY from the thread title? It still works, but it takes more effort than I had originally anticipated. But I really, really like the results and the control this gives you over the beat meter.

I'm keen to hear further advise on how to improve this, or suggestions on better ways to do accomplish this that a Fusion newb like me doesn't know exists. I want to say that conditionals will help here, but I don't yet grok how those work. Something like 'if ( NUMBERVALUE > .9) then NUMBERVALUE = 1 else NUMBERVALUE = 0' would do wonders. I think.

[masperturbator]

Suck Less is precise in its input WAV reading. The text node I tested will move exactly as directed by the modifier. The problem you found is in the particle emitter, and I don't see an obvious solution to get the emitter to behave that way. It might be that the particle Number value isn't entirely what we think it is.

This is speculation on my part, but I believe that Audacity might not be setting 'silence' to exactly 0dB. If true this will result in double-beats being generated from time to time, since the particle emitter generates particles based on probability, and 'nearly zero' probability is not nearly-zero enough.

The beat sample WAV being fed into BMG has no Audacity generated silence.



BMG doesn't add noise.



I reduced the BPM from 500 to 300 to confirm that the non-zero minimum value we see in the 500 BPM test is caused by sampling rate resolution. That means:



This slope that occurs can't be avoided at BPM near 500. At 300 BPM, Suck Less puts 1 and 0 in the Fusion modified field. At 500 BPM, it inputs 1 and a point on that slope before 0, but very close to zero.

Reducing the track to 300 BPM didn't change how the particle emitter behaves. The text Fusion comp is still very accurate in its response, and the particle comp is still as inaccurate.

[MortonBridges]

Having completed my newest Cock Hero, my first to to use this method, I can say that it works. Kind of. There's a good amount of tweaking and patching that's involved. Would I recommend it? Maybe not. Will I use it again? Probably. It gives me the complete control over how the beat bar functions, which I like, and I can live with the seemingly endless tweaking that's necessary. But life isn't difficult enough for me (heck, I do all my editing on Linux because I like the challenge) so your mileage may vary.

I'm getting a better handle on how to handle the tweaks, so my next effort might not be so frustrating. I'm still likely not doing things in the best way possible. As I discover better ways to do things with this I'll post here. And if anyone has any hints or suggestions I'd love to hear them.

[0088_]

So, I gave it a try and I actually succeeded in creating beat markers (and text) that are synced to a click track! I created the click track using Reaper using the built-in metronome, no conversion to 'square waves' necessary! I only needed to tweak the "amplitude scale" for the modifier in Davinci so that all clicks were generating a dot, very easy to do!

See the results in this link.
https://vimeo.com/manage/videos/828163959
I did not go through the effort of aligning the click track to the music, so the result sounds like shit, but the synced markers are there.