chrysaora 🎐
@chrysaoravioli.bsky.social
PoC "zone capping" feature like Splatoon.
Instead of using my central "Paint Stamper" system, the Zone system painted the zone. But, that means the Paint Stamper is unaware of the changes so I currently have a questionable workaround that fires every frame.
Instead of using my central "Paint Stamper" system, the Zone system painted the zone. But, that means the Paint Stamper is unaware of the changes so I currently have a questionable workaround that fires every frame.
November 20, 2025 at 4:02 AM
PoC "zone capping" feature like Splatoon.
Instead of using my central "Paint Stamper" system, the Zone system painted the zone. But, that means the Paint Stamper is unaware of the changes so I currently have a questionable workaround that fires every frame.
Instead of using my central "Paint Stamper" system, the Zone system painted the zone. But, that means the Paint Stamper is unaware of the changes so I currently have a questionable workaround that fires every frame.
I don't have any interesting breakdowns for the Niagara System since most of the code is very sloppy for now.
Instead, here's an image of my "version control".
I don't know how to set it up so I'm just packaging the plugin every week or so and zipping up the contents to a backup hard drive 🫡
Instead, here's an image of my "version control".
I don't know how to set it up so I'm just packaging the plugin every week or so and zipping up the contents to a backup hard drive 🫡
November 19, 2025 at 4:15 AM
I don't have any interesting breakdowns for the Niagara System since most of the code is very sloppy for now.
Instead, here's an image of my "version control".
I don't know how to set it up so I'm just packaging the plugin every week or so and zipping up the contents to a backup hard drive 🫡
Instead, here's an image of my "version control".
I don't know how to set it up so I'm just packaging the plugin every week or so and zipping up the contents to a backup hard drive 🫡
Hey, it's Splat Zones!
I can define any n-sided polygon and voila, we can detect paint in that area. Sampling resolution is kept low about 1px/m. Don't need to be too accurate.
Next, I'll introduce a fill operation, so when it hits 75%, it completely fills the zone with the winning team's color.
I can define any n-sided polygon and voila, we can detect paint in that area. Sampling resolution is kept low about 1px/m. Don't need to be too accurate.
Next, I'll introduce a fill operation, so when it hits 75%, it completely fills the zone with the winning team's color.
November 19, 2025 at 4:03 AM
Hey, it's Splat Zones!
I can define any n-sided polygon and voila, we can detect paint in that area. Sampling resolution is kept low about 1px/m. Don't need to be too accurate.
Next, I'll introduce a fill operation, so when it hits 75%, it completely fills the zone with the winning team's color.
I can define any n-sided polygon and voila, we can detect paint in that area. Sampling resolution is kept low about 1px/m. Don't need to be too accurate.
Next, I'll introduce a fill operation, so when it hits 75%, it completely fills the zone with the winning team's color.
I had an issue where "inward points" in a polygon would cause issues. Found this 2D Polygon SDF example on ShaderToy that solves my issue: www.shadertoy.com/view/wdBXRW
Tbh, I don't expect my detection zones to have any winding but I suppose it's a nice feature to have.
Tbh, I don't expect my detection zones to have any winding but I suppose it's a nice feature to have.
November 18, 2025 at 4:31 AM
I had an issue where "inward points" in a polygon would cause issues. Found this 2D Polygon SDF example on ShaderToy that solves my issue: www.shadertoy.com/view/wdBXRW
Tbh, I don't expect my detection zones to have any winding but I suppose it's a nice feature to have.
Tbh, I don't expect my detection zones to have any winding but I suppose it's a nice feature to have.
Buffing the long overdue Paint Detection side of the plugin.
Spent the weekend researching the math for creating an irregular n-sided polygon to replace the previous primitive cube method.
Doesn't work with all shapes right now (such as a star) but I'll continue debugging it tomorrow.
Spent the weekend researching the math for creating an irregular n-sided polygon to replace the previous primitive cube method.
Doesn't work with all shapes right now (such as a star) but I'll continue debugging it tomorrow.
November 17, 2025 at 6:04 AM
Buffing the long overdue Paint Detection side of the plugin.
Spent the weekend researching the math for creating an irregular n-sided polygon to replace the previous primitive cube method.
Doesn't work with all shapes right now (such as a star) but I'll continue debugging it tomorrow.
Spent the weekend researching the math for creating an irregular n-sided polygon to replace the previous primitive cube method.
Doesn't work with all shapes right now (such as a star) but I'll continue debugging it tomorrow.
Demo for Modular and Proxy Paint Actors.
Modular contains a singular visible mesh with a single corresponding collision mesh.
Proxy allows multiple visible meshes correspond to a single collision mesh, useful for large collision meshes where multiple static meshes may be associated with it.
Modular contains a singular visible mesh with a single corresponding collision mesh.
Proxy allows multiple visible meshes correspond to a single collision mesh, useful for large collision meshes where multiple static meshes may be associated with it.
November 8, 2025 at 5:26 AM
Demo for Modular and Proxy Paint Actors.
Modular contains a singular visible mesh with a single corresponding collision mesh.
Proxy allows multiple visible meshes correspond to a single collision mesh, useful for large collision meshes where multiple static meshes may be associated with it.
Modular contains a singular visible mesh with a single corresponding collision mesh.
Proxy allows multiple visible meshes correspond to a single collision mesh, useful for large collision meshes where multiple static meshes may be associated with it.
Remapped Atlas parameters are sent to the material via Custom Primitive Data.
Before, I was using Dynamic Material Instances but it's a pain to set up the code per material instance. Custom Primitive Data instead only needs one node that applies to every material that shares the same index!
Before, I was using Dynamic Material Instances but it's a pain to set up the code per material instance. Custom Primitive Data instead only needs one node that applies to every material that shares the same index!
November 6, 2025 at 3:27 AM
Remapped Atlas parameters are sent to the material via Custom Primitive Data.
Before, I was using Dynamic Material Instances but it's a pain to set up the code per material instance. Custom Primitive Data instead only needs one node that applies to every material that shares the same index!
Before, I was using Dynamic Material Instances but it's a pain to set up the code per material instance. Custom Primitive Data instead only needs one node that applies to every material that shares the same index!
For now, the collision mesh is separate from the visible mesh because Niagara builds the UV map from the collision mesh. Having a combined mesh means creating separate logic that I don't feel like implementing 😄
The paint UVs are fairly easy to create thanks to Blender's Data Transfer modifier.
The paint UVs are fairly easy to create thanks to Blender's Data Transfer modifier.
November 6, 2025 at 3:21 AM
For now, the collision mesh is separate from the visible mesh because Niagara builds the UV map from the collision mesh. Having a combined mesh means creating separate logic that I don't feel like implementing 😄
The paint UVs are fairly easy to create thanks to Blender's Data Transfer modifier.
The paint UVs are fairly easy to create thanks to Blender's Data Transfer modifier.
Added support for a "packed" paintable mesh.
If a static mesh is added to the "Visible Mesh" variable, it will automatically create a component to show the added mesh. Perfect for modular paintable actors similar to Splatoon 3's single player campaign.
Will work on detached visible meshes next...
If a static mesh is added to the "Visible Mesh" variable, it will automatically create a component to show the added mesh. Perfect for modular paintable actors similar to Splatoon 3's single player campaign.
Will work on detached visible meshes next...
November 6, 2025 at 3:13 AM
Added support for a "packed" paintable mesh.
If a static mesh is added to the "Visible Mesh" variable, it will automatically create a component to show the added mesh. Perfect for modular paintable actors similar to Splatoon 3's single player campaign.
Will work on detached visible meshes next...
If a static mesh is added to the "Visible Mesh" variable, it will automatically create a component to show the added mesh. Perfect for modular paintable actors similar to Splatoon 3's single player campaign.
Will work on detached visible meshes next...
Been a while since I've done a UE5 update.
The paint system generally works but I've decided to create a test map that forces me to flesh out various features.
Currently, I'm working on a system where a single paintable collision mesh to distribute painting params to corresponding visible meshes.
The paint system generally works but I've decided to create a test map that forces me to flesh out various features.
Currently, I'm working on a system where a single paintable collision mesh to distribute painting params to corresponding visible meshes.
November 5, 2025 at 4:38 AM
Been a while since I've done a UE5 update.
The paint system generally works but I've decided to create a test map that forces me to flesh out various features.
Currently, I'm working on a system where a single paintable collision mesh to distribute painting params to corresponding visible meshes.
The paint system generally works but I've decided to create a test map that forces me to flesh out various features.
Currently, I'm working on a system where a single paintable collision mesh to distribute painting params to corresponding visible meshes.
[ #NinjaGaiden4 Tech ]
Dodge Offset causes the final hit of the rapier's YYY combo to bug out. I don't even know if it's some weird root motion or motion warping error but it's neat to see that the hitbox is independent from the animation.
Dodge Offset causes the final hit of the rapier's YYY combo to bug out. I don't even know if it's some weird root motion or motion warping error but it's neat to see that the hitbox is independent from the animation.
October 25, 2025 at 3:04 AM
[ #NinjaGaiden4 Tech ]
Dodge Offset causes the final hit of the rapier's YYY combo to bug out. I don't even know if it's some weird root motion or motion warping error but it's neat to see that the hitbox is independent from the animation.
Dodge Offset causes the final hit of the rapier's YYY combo to bug out. I don't even know if it's some weird root motion or motion warping error but it's neat to see that the hitbox is independent from the animation.
[ #NinjaGaiden4 Tech ]
Using Dodge Offset, it's possible to infinitely loop between standard and Blood Raven combo strings.
In this example, I am looping between the dual swords' XXXXX string and the Blood Raven XXX string.
Using Dodge Offset, it's possible to infinitely loop between standard and Blood Raven combo strings.
In this example, I am looping between the dual swords' XXXXX string and the Blood Raven XXX string.
October 25, 2025 at 3:02 AM
[ #NinjaGaiden4 Tech ]
Using Dodge Offset, it's possible to infinitely loop between standard and Blood Raven combo strings.
In this example, I am looping between the dual swords' XXXXX string and the Blood Raven XXX string.
Using Dodge Offset, it's possible to infinitely loop between standard and Blood Raven combo strings.
In this example, I am looping between the dual swords' XXXXX string and the Blood Raven XXX string.
[ #NinjaGaiden4 Tech ]
By using Dodge Offset, you can retain your offset combo even after an Obliteration Technique.
By using Dodge Offset, you can retain your offset combo even after an Obliteration Technique.
October 25, 2025 at 3:02 AM
[ #NinjaGaiden4 Tech ]
By using Dodge Offset, you can retain your offset combo even after an Obliteration Technique.
By using Dodge Offset, you can retain your offset combo even after an Obliteration Technique.
This can also be combined with other weapons. Since all weapons have a Blood Raven XXY combo, I can loop their finishers.
October 23, 2025 at 4:34 PM
This can also be combined with other weapons. Since all weapons have a Blood Raven XXY combo, I can loop their finishers.
Weird Dodge Offset + BloodRaven infinite finisher (like Astral Chain sorta?)
Dual Sword Combo: XXXXY
Blood Raven Combo: XXY
Input Sequence
XXXX (Dual Sword) -> Dodge Offset -> Y (Blood Raven)
Just keeping using Dodge Offset to infinitely loop Blood Raven XXY.
#NinjaGaiden4
Dual Sword Combo: XXXXY
Blood Raven Combo: XXY
Input Sequence
XXXX (Dual Sword) -> Dodge Offset -> Y (Blood Raven)
Just keeping using Dodge Offset to infinitely loop Blood Raven XXY.
#NinjaGaiden4
October 23, 2025 at 4:34 PM
Weird Dodge Offset + BloodRaven infinite finisher (like Astral Chain sorta?)
Dual Sword Combo: XXXXY
Blood Raven Combo: XXY
Input Sequence
XXXX (Dual Sword) -> Dodge Offset -> Y (Blood Raven)
Just keeping using Dodge Offset to infinitely loop Blood Raven XXY.
#NinjaGaiden4
Dual Sword Combo: XXXXY
Blood Raven Combo: XXY
Input Sequence
XXXX (Dual Sword) -> Dodge Offset -> Y (Blood Raven)
Just keeping using Dodge Offset to infinitely loop Blood Raven XXY.
#NinjaGaiden4
Some more screenshots. Loving the PC version so far.
It'll be pretty interesting to compare this to the Ninja Gaiden 2 Black. I do feel that NG2B's lighting looks "better" but I appreciate that NG4's look is more stable and avoids some of the gathering artifacts that Lumen occasionally has.
It'll be pretty interesting to compare this to the Ninja Gaiden 2 Black. I do feel that NG2B's lighting looks "better" but I appreciate that NG4's look is more stable and avoids some of the gathering artifacts that Lumen occasionally has.
October 22, 2025 at 5:17 AM
Some more screenshots. Loving the PC version so far.
It'll be pretty interesting to compare this to the Ninja Gaiden 2 Black. I do feel that NG2B's lighting looks "better" but I appreciate that NG4's look is more stable and avoids some of the gathering artifacts that Lumen occasionally has.
It'll be pretty interesting to compare this to the Ninja Gaiden 2 Black. I do feel that NG2B's lighting looks "better" but I appreciate that NG4's look is more stable and avoids some of the gathering artifacts that Lumen occasionally has.
Amazon delayed my physical copy for PS5 to Friday so... I guess I'll double dip just this once.
Not usually the biggest PC guy but hey, it looks pretty nice on my ultrawide. Could use a new graphics card though because it's struggling a bit (plus I hate fiddling with settings).
Not usually the biggest PC guy but hey, it looks pretty nice on my ultrawide. Could use a new graphics card though because it's struggling a bit (plus I hate fiddling with settings).
October 21, 2025 at 8:06 PM
Amazon delayed my physical copy for PS5 to Friday so... I guess I'll double dip just this once.
Not usually the biggest PC guy but hey, it looks pretty nice on my ultrawide. Could use a new graphics card though because it's struggling a bit (plus I hate fiddling with settings).
Not usually the biggest PC guy but hey, it looks pretty nice on my ultrawide. Could use a new graphics card though because it's struggling a bit (plus I hate fiddling with settings).
Hit masking using a single channel is somewhat working! Still a few bugs to work out but it works a majority of the time.
Hit a wall? Then it only affects the surface at that particular angle.
Hit a floor? Then it only affects floors within a certain depth like Splatoon.
Hit a wall? Then it only affects the surface at that particular angle.
Hit a floor? Then it only affects floors within a certain depth like Splatoon.
October 19, 2025 at 10:39 PM
Hit masking using a single channel is somewhat working! Still a few bugs to work out but it works a majority of the time.
Hit a wall? Then it only affects the surface at that particular angle.
Hit a floor? Then it only affects floors within a certain depth like Splatoon.
Hit a wall? Then it only affects the surface at that particular angle.
Hit a floor? Then it only affects floors within a certain depth like Splatoon.
Hey, actually not too hard!
Converted the entirety of the wall normalization code from the material graph into Niagara using a single HLSL node.
Now, I've got perfectly packed Floor depth values and Wall normals in the alpha channel of my position map. On to depth masking experimentation!
Converted the entirety of the wall normalization code from the material graph into Niagara using a single HLSL node.
Now, I've got perfectly packed Floor depth values and Wall normals in the alpha channel of my position map. On to depth masking experimentation!
October 15, 2025 at 3:40 AM
Hey, actually not too hard!
Converted the entirety of the wall normalization code from the material graph into Niagara using a single HLSL node.
Now, I've got perfectly packed Floor depth values and Wall normals in the alpha channel of my position map. On to depth masking experimentation!
Converted the entirety of the wall normalization code from the material graph into Niagara using a single HLSL node.
Now, I've got perfectly packed Floor depth values and Wall normals in the alpha channel of my position map. On to depth masking experimentation!
[ Niagara Depth Masking ]
Porting over my material graph code into Niagara, piece by piece.
Not too much time tonight. Focused only on code that determines which parts of the paint map are considered floors vs. walls. These outputs will be used later in the script for compositing purposes.
Porting over my material graph code into Niagara, piece by piece.
Not too much time tonight. Focused only on code that determines which parts of the paint map are considered floors vs. walls. These outputs will be used later in the script for compositing purposes.
October 14, 2025 at 3:14 AM
[ Niagara Depth Masking ]
Porting over my material graph code into Niagara, piece by piece.
Not too much time tonight. Focused only on code that determines which parts of the paint map are considered floors vs. walls. These outputs will be used later in the script for compositing purposes.
Porting over my material graph code into Niagara, piece by piece.
Not too much time tonight. Focused only on code that determines which parts of the paint map are considered floors vs. walls. These outputs will be used later in the script for compositing purposes.
Next, I'll recreate the math seen here in Niagara and then hopefully, can finally begin creating a masking solution.
In this image, I only want the splatter to apply to the wall but it's showing on some undesired locations (wall side & floor). The masking solution should be able to remedy this. 3/🧵
In this image, I only want the splatter to apply to the wall but it's showing on some undesired locations (wall side & floor). The masking solution should be able to remedy this. 3/🧵
October 13, 2025 at 1:54 AM
Next, I'll recreate the math seen here in Niagara and then hopefully, can finally begin creating a masking solution.
In this image, I only want the splatter to apply to the wall but it's showing on some undesired locations (wall side & floor). The masking solution should be able to remedy this. 3/🧵
In this image, I only want the splatter to apply to the wall but it's showing on some undesired locations (wall side & floor). The masking solution should be able to remedy this. 3/🧵
When paint hits a floor, I allow it to be shown in a certain range like Splatoon.
Allowed range is (-65504 <= z < -1) & (1 < z <= 65504). For simplicity, my paintable floors are placed within a range of (100 < z < 65504).
In this example, I'm only showing all floors that are located at 800cm.
2/🧵
Allowed range is (-65504 <= z < -1) & (1 < z <= 65504). For simplicity, my paintable floors are placed within a range of (100 < z < 65504).
In this example, I'm only showing all floors that are located at 800cm.
2/🧵
October 13, 2025 at 1:48 AM
When paint hits a floor, I allow it to be shown in a certain range like Splatoon.
Allowed range is (-65504 <= z < -1) & (1 < z <= 65504). For simplicity, my paintable floors are placed within a range of (100 < z < 65504).
In this example, I'm only showing all floors that are located at 800cm.
2/🧵
Allowed range is (-65504 <= z < -1) & (1 < z <= 65504). For simplicity, my paintable floors are placed within a range of (100 < z < 65504).
In this example, I'm only showing all floors that are located at 800cm.
2/🧵
Packing RGB Normal Map into a single channel so I can potentially halve the compute time for paint splatters.
When a paint splatter hits a wall, the hit angle is from 0 to 360 degrees, which can be normalized to a value from -1 to 1 (think unit circle).
1/🧵
When a paint splatter hits a wall, the hit angle is from 0 to 360 degrees, which can be normalized to a value from -1 to 1 (think unit circle).
1/🧵
October 13, 2025 at 1:36 AM
Packing RGB Normal Map into a single channel so I can potentially halve the compute time for paint splatters.
When a paint splatter hits a wall, the hit angle is from 0 to 360 degrees, which can be normalized to a value from -1 to 1 (think unit circle).
1/🧵
When a paint splatter hits a wall, the hit angle is from 0 to 360 degrees, which can be normalized to a value from -1 to 1 (think unit circle).
1/🧵
The plan is to store normalized wall angles to a range of -1 to 1, while the floor z-height can be a range of 100 to 65504.
So depending on what sort of splatter you have (floor vs. wall), I can just use this single channel for masking and save on both memory (bye extra 64mb RT) and compute time.
So depending on what sort of splatter you have (floor vs. wall), I can just use this single channel for masking and save on both memory (bye extra 64mb RT) and compute time.
September 29, 2025 at 2:32 AM
The plan is to store normalized wall angles to a range of -1 to 1, while the floor z-height can be a range of 100 to 65504.
So depending on what sort of splatter you have (floor vs. wall), I can just use this single channel for masking and save on both memory (bye extra 64mb RT) and compute time.
So depending on what sort of splatter you have (floor vs. wall), I can just use this single channel for masking and save on both memory (bye extra 64mb RT) and compute time.
Had a Sunday night epiphany.
Before, I was doing two texture samples to mask a paint splatter but this doubles the compute time. It'd be nice if I could just use a single texture, so I'm trying to cram the normal map info into the unused alpha channel of my World Position texture.
Before, I was doing two texture samples to mask a paint splatter but this doubles the compute time. It'd be nice if I could just use a single texture, so I'm trying to cram the normal map info into the unused alpha channel of my World Position texture.
September 29, 2025 at 2:22 AM
Had a Sunday night epiphany.
Before, I was doing two texture samples to mask a paint splatter but this doubles the compute time. It'd be nice if I could just use a single texture, so I'm trying to cram the normal map info into the unused alpha channel of my World Position texture.
Before, I was doing two texture samples to mask a paint splatter but this doubles the compute time. It'd be nice if I could just use a single texture, so I'm trying to cram the normal map info into the unused alpha channel of my World Position texture.