What's new

Recreating Edge Flare in Resolve

Ryley Konechny

Ryley Konechny

Hello! I was wondering if anyone has attempted to recreate an edge flare effect. For those who don't know it has something to do with light just outside the lens or camera causing flaring around the edges of the frame. I've noticed it primarily in film and some digital.

I've tried to recreate this effect but haven't had that much luck. I know it can be done in Nuke as seen in the links provided, but I don't quite know my way around it yet.

I would love to hear what you guys have to say about this effect and if anyone has any ways to achieve it!

https://www.chrisbturner.com/blog/creating-a-2d-point-based-procedural-edge-flare-tool-in-nuke

http://www.nukepedia.com/toolsets/draw/ct_edgeflare

Also try to stay on subject. These things tend to stray off onto other topics that don't pertain to the main question at all and I don't really care why someone shouldn't do this or how they shouldn't do this. That type of discussion doesn't help with anything.
 
I had to emulate a film projector beam on a theater setup, and I masked around the actors head and tracked it so the beam would stay mostly on the edges.

you could try that with light flares, allowing it to wrap around the edges of mask a bit.
 
Ryley Konechny said:
Hello! I was wondering if anyone has attempted to recreate an edge flare effect. For those who don't know it has something to do with light just outside the lens or camera causing flaring around the edges of the frame. I've noticed it primarily in film and some digital.
Click to expand...
Can you post a photo of the effect you're trying to create? Halation and Glow are readily available in most color software these days. There is something very real about using glass filters in camera for flare and glow, but I think the OFX Scatter plug-in can get pretty close to that.
 
To do it properly, you'd definitely need to get into Nuke or Fusion. Chris (who made the Nuke demo in your link) is a colleague of mine - I fully endorse his techniques.

I just had a go at building an extremely rudimentary effect of that nature in Resolve. It totally lacks the kind of texture and breakup and complexity needed, but it might roughly do what you need in certain situations, especially if you play it quite subtle in the mix.

Here's a quick look at what it does -

And you can download the powergrade here:

https://drive.google.com/file/d/1wgkqUO6bZOrl9gSFmj4EMY14td3l9KBm/view?usp=sharing

It's a compound node that expects (and outputs) LogC AWG, so colourspace transform in and out of LogC AWG either side of the node if needed. Here's what you'll find inside the compound node:

edgeflare_compound.jpg

Node 02 - Threshold. Use Gain up/down to increase/lower the sensitivity of the input. i.e. higher gain will push more values into the effect's catchment range, lower values will decrease how much brightness gets picked up.

Node 16 - Use Gain to increase/Decrease the intensity/opacity of the result.

A couple other things to bear in mind:

- The mattes are optimised for 16:9 footage (and 16:9 timeline) - you'll need to adjust the Outer and Inner mattes on the X branch if your footage is in a different aspect. Try it on 16:9 footage first to see where the window mask edges should end up relative to frame, so you can tweak accordingly for anything else.

- This setup will add ~5% upscale to the footage. This is to ensure the flare starts to kick into frame while the source is still just out of frame. This wouldn't be possible otherwise. The more you increase the node scaling in Node 8, the more delay there will be between flare appearing and source appearing into frame.

Have a play - plenty of ways to tweak it I'm sure. Sorry if anything breaks - haven't tested it much!
 
Last edited:
Miguel Santana said:
To do it properly, you'd definitely need to get into Nuke or Fusion. Chris (who made the Nuke demo in your link) is a colleague of mine - I fully endorse his techniques.

But I just had a go at building an extremely rudimentary effect of that nature in Resolve. It totally lacks the kind of texture and breakup and complexity needed, but it might roughly do what you need in certain situations, especially if you play it quite subtle in the mix.

Here's a quick look at what it does -

And you can download the powergrade here:

https://drive.google.com/file/d/1wgkqUO6bZOrl9gSFmj4EMY14td3l9KBm/view?usp=sharing

It's a compound node that expects (and outputs) LogC AWG, so colourspace transform in and out of LogC AWG either side of the node if needed. Here's what you'll find inside the compound node:

View attachment 9936

Node 02 - Threshold. Use Gain up/down to increase/lower the sensitivity of the input. i.e. higher gain will push more values into the effect's catchment range, lower values will decrease how much brightness gets picked up.

Node 16 - Use Gain to increase/Decrease the intensity/opacity of the result.

A couple other things to bear in mind:

- The mattes are optimised for 16:9 footage (and 16:9 timeline) - you'll need to adjust the Outer and Inner mattes on the X branch if your footage is in a different aspect. Try it on 16:9 footage first to see where the window mask edges should end up relative to frame, so you can tweak accordingly for anything else.

- This setup will add ~5% upscale to the footage. This is to ensure the flare starts to kick into frame while the source is still just out of frame. This wouldn't be possible otherwise.

Have a play - plenty of ways to tweak it I'm sure. Sorry if anything breaks - haven't tested it much!
Click to expand...
Wow thank you so much Miguel, really appreciate it!
 
Miguel Santana said:
To do it properly, you'd definitely need to get into Nuke or Fusion. Chris (who made the Nuke demo in your link) is a colleague of mine - I fully endorse his techniques.

I just had a go at building an extremely rudimentary effect of that nature in Resolve. It totally lacks the kind of texture and breakup and complexity needed, but it might roughly do what you need in certain situations, especially if you play it quite subtle in the mix.

Here's a quick look at what it does -

And you can download the powergrade here:

https://drive.google.com/file/d/1wgkqUO6bZOrl9gSFmj4EMY14td3l9KBm/view?usp=sharing

It's a compound node that expects (and outputs) LogC AWG, so colourspace transform in and out of LogC AWG either side of the node if needed. Here's what you'll find inside the compound node:

View attachment 9936

Node 02 - Threshold. Use Gain up/down to increase/lower the sensitivity of the input. i.e. higher gain will push more values into the effect's catchment range, lower values will decrease how much brightness gets picked up.

Node 16 - Use Gain to increase/Decrease the intensity/opacity of the result.

A couple other things to bear in mind:

- The mattes are optimised for 16:9 footage (and 16:9 timeline) - you'll need to adjust the Outer and Inner mattes on the X branch if your footage is in a different aspect. Try it on 16:9 footage first to see where the window mask edges should end up relative to frame, so you can tweak accordingly for anything else.

- This setup will add ~5% upscale to the footage. This is to ensure the flare starts to kick into frame while the source is still just out of frame. This wouldn't be possible otherwise. The more you increase the node scaling in Node 8, the more delay there will be between flare appearing and source appearing into frame.

Have a play - plenty of ways to tweak it I'm sure. Sorry if anything breaks - haven't tested it much!
Click to expand...

Thanks for sharing this, Miguel. Lovely solution. It's very refreshing to see the color nodes used in such a way.
 
I'm currently working on a node that uses Chris Turner's method on 2D imagery. I think his node is intended more for 3D VFX work with a virtual camera and self-generated light spots, but to be honest, I am not sure. There is probably a way to simply use his node on 2D imagery, but looking at how this might work and breaking down node trees in this manner usually overwhelms me. The two issues, to my knowledge, are that his node works on producing lens flares based on a singular pixel value position, and I'm not sure yet how I would deal with this when there are multiple sources of lights that need to flare up.

Intuitively, that would mean the node needs to be able to discern a way to only have a singular flare pixel for each light source, but how can it easily differentiate between them?

Secondly, I am really surprised you mentioned this happening on digital; I thought this was an effect only visible in analog film. I think it has more to do with the film gate than the lens, but again, that's my assumption. The issue is that from the examples I have seen, these flares seem to be very specific to extremely hot sources. The problem I find with digital is that with its highlight latitude, it is hard to find a threshold in linear, which only affects these super bright sources. I have a similar issue with trying to achieve the strong halo halation also often seen in analog film. I haven't figured out how to differentiate sources that produce the large halo from those that should normally halate. Though perhaps a highly EI-rated Alexa 35 or Venice 2 may allow for this.

www.yedlin.net

DISPLAY PREP DEMO v2

www.yedlin.net www.yedlin.net

This type of flare can be seen in Steve Yedlin's ASC display prep demo (06:39), and it might actually be the giveaway for which footage is the film.
 
Silvan Schnelli said:
I'm currently working on a node that uses Chris Turner's method on 2D imagery. I think his node is intended more for 3D VFX work with a virtual camera and self-generated light spots, but to be honest, I am not sure. There is probably a way to simply use his node on 2D imagery, but looking at how this might work and breaking down node trees in this manner usually overwhelms me. The two issues, to my knowledge, are that his node works on producing lens flares based on a singular pixel value position, and I'm not sure yet how I would deal with this when there are multiple sources of lights that need to flare up.

Intuitively, that would mean the node needs to be able to discern a way to only have a singular flare pixel for each light source, but how can it easily differentiate between them?

Secondly, I am really surprised you mentioned this happening on digital; I thought this was an effect only visible in analog film. I think it has more to do with the film gate than the lens, but again, that's my assumption. The issue is that from the examples I have seen, these flares seem to be very specific to extremely hot sources. The problem I find with digital is that with its highlight latitude, it is hard to find a threshold in linear, which only affects these super bright sources. I have a similar issue with trying to achieve the strong halo halation also often seen in analog film. I haven't figured out how to differentiate sources that produce the large halo from those that should normally halate. Though perhaps a highly EI-rated Alexa 35 or Venice 2 may allow for this.

www.yedlin.net

DISPLAY PREP DEMO v2

www.yedlin.net www.yedlin.net

This type of flare can be seen in Steve Yedlin's ASC display prep demo (06:39), and it might actually be the giveaway for which footage is the film.
Click to expand...
Looking forward to seeing what you can do, Silvan and would love to try it out if you need any testing!
 
You must log in or register to reply here.
Back
Top