Convolution Bloom Kernels Pack

Create beautiful looking Bloom/Glare effects using Convolution Bloom Kernel system built in directly in Unreal Engine. Now you can use real calculated bloom via convolution and FFT instead of lens flare textures as emissive planes.

Showcase Video: https://youtu.be/r2kS1CxUBm8

Quick tips on how to use:

Set the bloom type to convolution in PPV, from this point onwards, all standard bloom settings are obsolete.

• Start with "Convolution Scale" set to 0.5. This pack includes kernels with twice the size of default one, so that you can increase the size too if you want. It is very useful to play with the scale to get the look you want.
• "Convolution Center" > Every bloom is set perfectly in the center, you do not need to adjust this at all.
• Convolution Boost Min, Max and the Multiplier are the key to use the bloom kernel. If you set the multiplier to one, It is not going to boost any pixels' intensity before going in to the bloom. So every pixel on the screen is boosted relative to their luminance. Typically it is better to use the other two, than to change the Convolution Bloom Max value. So as a general example, Convolution Boost Min looks good from 0.2 to 2 but it is highly highly dependent on the scene and lighting. It is same with Boost Multiplier, which can be usually set around 2 to 60. It is also possible to set this value lower than 1, which in that case it will reduce the intensity of the selected pixel brightness levels. (somewhat similar to a highlight compression)

If you want a controlled singular point bloom highlight at any surface, you can put a very small sphere(i.e. 0.02 size) with an emissive texture.

The bloom characteristics are highly dependent on your scene lighting, and every bloom is designed for a different scenario. And since it is a pixel based effect, screen size also changes how it looks.

Update:

• Added Support for UE 5.0
• Added 2 New Kernels; Film Halation and Festive
• Tweaked the falloff of some kernels to be more usable at dim light situations