Using textures for Bokehs

Hello fellow flamers. Is there a way we can inject textures to customize our bokehs in our blurs?

Thanks in advanced!

Yeah, the 3d Blur in Batch will take a kernel.

There’s some way to get the Action Matchboxes Physical Bokeh and Physical Defocus to do it, but I don’t have anything nice to say about how those nodes behave or render, so I suggest you use the 3d Blur. It’s slow, but you’ll get what you’re looking for.

Also Sapphire Convolve does this.


Yea, what Andy said. Unfortunately the MX “Physical” features feel half developed and I haven’t been able to use them in production. So just use the old, slow, but effective tool.

awesome! thanks guys!

Actually I’d try Sapphire first, because I think they update the tools. 3D Blur - it’s been a minute or two.

1 Like

I’m curious if anyone uses any of the MX “physical” tools. It had great potential, but always felt like an alpha release that no one uses. I tried and tried, but I am not smart enough to make it work well and easy.

Why release things like this, then just move on to release other half-tools? Sorry, this is a persistent complaint with Flame development. Tremendous potential, half-developed, then forgotten.

Probably 10 other features in that category too.

It is so frustrating… They spent their dev budget doing it wrong the first time around, and then it’s just hack patching on top to try and make it work. MotionVectorTracking is one of the prime examples of this currently. The original approach was just totally wrong and broken. They’ve been able to improve it a bit, bit still many failure scenarios and the only path forward is to scrap it all and re-write, but they can’t because they shot the budget on that years ago. Slow death.

1 Like

Could have been a game changer. Dropped to pursue other features. Most no one knows about the feature because it wasn’t well promoted, and more importantly, it wasn’t given enough time and resources to develop properly.


If I can put my fake “business expert” hat on for a moment1, these tools were developed to tick a box off both in the “Longstanding Feature Request” pile and the “Marketing wants something they can sell” pile, and were thus built.

Once that happens, you have the insta-freeze of all flame features2 (shout out to the ten years without “drag” in batch paint), and I would assume the dev team is pushed by the business experts at Autodesk to make new stuff. In addition to all this, the nodes are matchboxes buried inside of Action, which i’m guessing means only (charitably) 20% of users even knew the tools existed. This leads to minimal dev feedback and a presumption that the nodes aren’t being used so why invest in them further. A loop of apathy.

You can lump a LOT of tools into this pile, from the old camera tracker, to the distort node, to the new camera tracker, to action particles, and so on.

1. I am not a business expert. My only economically valuable skill is driving a flame, and I commend the devs on all their tireless hard work keeping people like me both happy and employed. Were I put in charge of flame development most of this forum would hate me within a week.
2. One great feature of matchbox-based tools is they load the entire tool into the setup, so should Autodesk completely overhaul your favorite matchbox node, the one in your old setup will still render identically since it’s the old code. Now, I think this has gone too far and become the only option for new visual tools in flame, which is problematic both because of the limits of a GLSL shader and because of how buried matchbox nodes are in the software (Master Grade should be a node, for example), further liminting user adoption and thus dev feedback.

Yes… but if they are going to do it, do it right the first time, so that they don’t end up in the situation that there isn’t money to have to try and make it right the second or third time.